For decades, Delaware has brought together scientific research and advanced manufacturing to address practical challenges. The First State has developed a strong track record of turning research breakthroughs into technologies with applications far beyond its borders.
The United States is now competing globally for leadership in artificial intelligence (AI), and maintaining that position will require a rapid expansion of next-generation data center infrastructure. Much of the discussion surrounding this expansion has centered on securing additional electricity generation and water supplies. However, an equally important question is how efficiently those resources are used once they reach a data center.
A substantial portion of the electricity consumed by data centers goes toward something other than computing. It is needed to remove the enormous amounts of heat produced by increasingly powerful processors. More than 95% of data centers currently use conventional air-cooling systems, while cooling IT equipment can account for more than 40% of a facility’s overall energy consumption. Cooling infrastructure can also require considerable quantities of water.
As demand for AI, cloud computing and high-performance computing grows, simply obtaining more electricity and water is no longer enough. Data center operators also need to maximize the computing value generated from every unit of energy and make more efficient use of available water resources.
The issue is becoming increasingly difficult as AI processors become more powerful. New chip generations deliver higher levels of computing performance, but they also produce greater amounts of heat. As a result, industry specialists are increasingly examining the limitations of conventional air cooling and even single-phase liquid-cooling approaches when applied to the latest high-performance hardware.
A new generation of cooling innovation is emerging in Delaware to address this challenge.
At the Chemours Discovery Hub (CDH) on the University of Delaware’s STAR Campus in Newark, more than 300 scientists, engineers and researchers are working on technologies intended to address emerging industrial and societal challenges. Among their areas of work are advanced liquid-cooling solutions aimed at improving the energy and water efficiency of data centers.
Traditional air cooling uses large chillers and fans to circulate cooled air throughout a facility. Liquid cooling takes a different approach by transferring heat directly from its source. Chemours’ Opteon™ two-phase direct-to-chip fluids, which are currently available, deliver cooling directly to high-heat-generating chips inside servers and can significantly improve cooling efficiency.
Chemours is also developing Opteon™ two-phase immersion-cooling technologies for future AI infrastructure. With this approach, servers are housed inside sealed containers and immersed in Opteon™ two-phase immersion-cooling fluid. When the equipment generates heat, the fluid boils and turns into vapor, carrying heat away from the servers. The vapor subsequently condenses back into liquid, producing a closed-loop cooling process.
This technology represents a different way of thinking about resource efficiency in data centers. Compared with conventional air-cooling systems, two-phase immersion cooling can reduce cooling-related energy consumption by as much as 90% and can nearly eliminate water consumption. It may also lower noise generated by cooling fans and reduce the physical footprint required for data center facilities.
One reason for these potential gains is that the liquid captures heat from the entire system rather than focusing only on the components that generate the most heat, such as central processing units (CPUs) and graphics processing units (GPUs).
The implications extend beyond cooling technology itself. The future expansion of AI will depend not only on the amount of electricity generation and water infrastructure that can be developed, but also on how effectively those resources are utilized. Electricity saved through more efficient cooling can instead support additional computing capacity, while water conservation can ease demands on local supplies.
As AI adoption continues to expand, improving the efficiency of existing energy and water resources will be an important part of building sustainable digital infrastructure. Through technologies such as two-phase direct-to-chip and immersion cooling, researchers and engineers in Delaware are working toward data centers that can deliver greater computing power while using resources more efficiently.
The United States is now competing globally for leadership in artificial intelligence (AI), and maintaining that position will require a rapid expansion of next-generation data center infrastructure. Much of the discussion surrounding this expansion has centered on securing additional electricity generation and water supplies. However, an equally important question is how efficiently those resources are used once they reach a data center.
A substantial portion of the electricity consumed by data centers goes toward something other than computing. It is needed to remove the enormous amounts of heat produced by increasingly powerful processors. More than 95% of data centers currently use conventional air-cooling systems, while cooling IT equipment can account for more than 40% of a facility’s overall energy consumption. Cooling infrastructure can also require considerable quantities of water.
As demand for AI, cloud computing and high-performance computing grows, simply obtaining more electricity and water is no longer enough. Data center operators also need to maximize the computing value generated from every unit of energy and make more efficient use of available water resources.
The issue is becoming increasingly difficult as AI processors become more powerful. New chip generations deliver higher levels of computing performance, but they also produce greater amounts of heat. As a result, industry specialists are increasingly examining the limitations of conventional air cooling and even single-phase liquid-cooling approaches when applied to the latest high-performance hardware.
A new generation of cooling innovation is emerging in Delaware to address this challenge.
At the Chemours Discovery Hub (CDH) on the University of Delaware’s STAR Campus in Newark, more than 300 scientists, engineers and researchers are working on technologies intended to address emerging industrial and societal challenges. Among their areas of work are advanced liquid-cooling solutions aimed at improving the energy and water efficiency of data centers.
Traditional air cooling uses large chillers and fans to circulate cooled air throughout a facility. Liquid cooling takes a different approach by transferring heat directly from its source. Chemours’ Opteon™ two-phase direct-to-chip fluids, which are currently available, deliver cooling directly to high-heat-generating chips inside servers and can significantly improve cooling efficiency.
Chemours is also developing Opteon™ two-phase immersion-cooling technologies for future AI infrastructure. With this approach, servers are housed inside sealed containers and immersed in Opteon™ two-phase immersion-cooling fluid. When the equipment generates heat, the fluid boils and turns into vapor, carrying heat away from the servers. The vapor subsequently condenses back into liquid, producing a closed-loop cooling process.
This technology represents a different way of thinking about resource efficiency in data centers. Compared with conventional air-cooling systems, two-phase immersion cooling can reduce cooling-related energy consumption by as much as 90% and can nearly eliminate water consumption. It may also lower noise generated by cooling fans and reduce the physical footprint required for data center facilities.
One reason for these potential gains is that the liquid captures heat from the entire system rather than focusing only on the components that generate the most heat, such as central processing units (CPUs) and graphics processing units (GPUs).
The implications extend beyond cooling technology itself. The future expansion of AI will depend not only on the amount of electricity generation and water infrastructure that can be developed, but also on how effectively those resources are utilized. Electricity saved through more efficient cooling can instead support additional computing capacity, while water conservation can ease demands on local supplies.
As AI adoption continues to expand, improving the efficiency of existing energy and water resources will be an important part of building sustainable digital infrastructure. Through technologies such as two-phase direct-to-chip and immersion cooling, researchers and engineers in Delaware are working toward data centers that can deliver greater computing power while using resources more efficiently.